| 摘要: |
| 基于大脑导航神经细胞机理的类脑认知地图构建方法，为发展智能同步定位与地图构建（SLAM）技术提供了新思路。针对现有类脑认知地图构建精度不高的问题，提出了一种基于双目视觉的类脑三维认知地图构建方法。首先阐述了类脑三维认知地图系统的工作原理，然后论述了不同视觉里程计对认知地图精度的影响，研究了基于双目视觉里程计的类脑三维认知地图精度优化方法，最终完成了基于视觉数据集的类脑三维认知地图构建试验。试验结果表明，所提方法构建的视觉里程计地图的三维位置误差为总行程的2.14%，认知地图的三维位置误差为1.56%；认知地图精度与里程计精度呈正相关；系统通过模板匹配进行回环检测与校正，提高了认知地图的精度。 |
| 关键词: 类脑导航 认知地图 SLAM 吸引子神经网络 视觉里程计 |
| DOI： |
|
| 基金项目:国家自然科学基金（61873125, 61673208, 61703208, 61533008, 61533009, 61973160）；江苏省自然科学基金（BK20181291）；上海航天科技创新基金（SAST2019-085）；中央高校基本科研业务费专项基金（NZ2019007） |
|
| A Method of Constructing Brain-inspired 3D Cognitive Maps Based on Binocular Vision |
| WANG Ya-ting,LIU Jian-ye,XIONG Zhi,YANG Chuang |
| (Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China;Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China; Key Laboratory of Navigation, Guidance and Health-Management Technologies of Advanced Aerocraft, Ministry of Industry and Information Technology, Nanjing 211106, China) |
| Abstract: |
| The construction methods of brain-inspired cognitive maps based on the mechanism of navigation neural cells provide new ideas for the development of intelligent simultaneous localization and mapping. Aiming at the problem of the low precision of existing methods, a method of constructing brain-inspired 3D cognitive maps based on binocular vision is proposed in this paper. This paper firstly expounds the operating principle of the brain-inspired 3D cognitive map system, then discusses the influence of different visual odometry on the accuracy of cognitive maps, and studies the accuracy optimization method of cognitive maps based on binocular visual odometry， and finally completes the brain-inspired cognitive maps constructing experiment based on visual datasets. The results show that the 3D position error of visual odometry maps is 2.14% of the total itinerary, the 3D position error of cognitive maps is 1.56%, the accuracy of cognitive maps is positively correlated with odometer accuracy， and the system improves the accuracy of cognitive maps by using template matching for loop detection and correction. |
| Key words: Brain-inspired navigation Cognitive maps SLAM Attractor neural network Visual odometry |
